diff --git a/.yarn/patches/@ai-sdk-google-npm-2.0.43-689ed559b3.patch b/.yarn/patches/@ai-sdk-google-npm-2.0.49-84720f41bd.patch
similarity index 64%
rename from .yarn/patches/@ai-sdk-google-npm-2.0.43-689ed559b3.patch
rename to .yarn/patches/@ai-sdk-google-npm-2.0.49-84720f41bd.patch
index 3015e702e..67403a657 100644
--- a/.yarn/patches/@ai-sdk-google-npm-2.0.43-689ed559b3.patch
+++ b/.yarn/patches/@ai-sdk-google-npm-2.0.49-84720f41bd.patch
@@ -1,5 +1,5 @@
 diff --git a/dist/index.js b/dist/index.js
-index 51ce7e423934fb717cb90245cdfcdb3dae6780e6..0f7f7009e2f41a79a8669d38c8a44867bbff5e1f 100644
+index d004b415c5841a1969705823614f395265ea5a8a..6b1e0dad4610b0424393ecc12e9114723bbe316b 100644
 --- a/dist/index.js
 +++ b/dist/index.js
 @@ -474,7 +474,7 @@ function convertToGoogleGenerativeAIMessages(prompt, options) {
@@ -12,7 +12,7 @@ index 51ce7e423934fb717cb90245cdfcdb3dae6780e6..0f7f7009e2f41a79a8669d38c8a44867
  
  // src/google-generative-ai-options.ts
 diff --git a/dist/index.mjs b/dist/index.mjs
-index f4b77e35c0cbfece85a3ef0d4f4e67aa6dde6271..8d2fecf8155a226006a0bde72b00b6036d4014b6 100644
+index 1780dd2391b7f42224a0b8048c723d2f81222c44..1f12ed14399d6902107ce9b435d7d8e6cc61e06b 100644
 --- a/dist/index.mjs
 +++ b/dist/index.mjs
 @@ -480,7 +480,7 @@ function convertToGoogleGenerativeAIMessages(prompt, options) {
@@ -24,3 +24,14 @@ index f4b77e35c0cbfece85a3ef0d4f4e67aa6dde6271..8d2fecf8155a226006a0bde72b00b603
  }
  
  // src/google-generative-ai-options.ts
+@@ -1909,8 +1909,7 @@ function createGoogleGenerativeAI(options = {}) {
+ }
+ var google = createGoogleGenerativeAI();
+ export {
+-  VERSION,
+   createGoogleGenerativeAI,
+-  google
++  google, VERSION
+ };
+ //# sourceMappingURL=index.mjs.map
+\ No newline at end of file
diff --git a/package.json b/package.json
index b894030b5..a70663ffc 100644
--- a/package.json
+++ b/package.json
@@ -114,8 +114,8 @@
     "@ai-sdk/anthropic": "^2.0.49",
     "@ai-sdk/cerebras": "^1.0.31",
     "@ai-sdk/gateway": "^2.0.15",
-    "@ai-sdk/google": "patch:@ai-sdk/google@npm%3A2.0.43#~/.yarn/patches/@ai-sdk-google-npm-2.0.43-689ed559b3.patch",
-    "@ai-sdk/google-vertex": "^3.0.79",
+    "@ai-sdk/google": "patch:@ai-sdk/google@npm%3A2.0.49#~/.yarn/patches/@ai-sdk-google-npm-2.0.49-84720f41bd.patch",
+    "@ai-sdk/google-vertex": "^3.0.94",
     "@ai-sdk/huggingface": "^0.0.10",
     "@ai-sdk/mistral": "^2.0.24",
     "@ai-sdk/openai": "patch:@ai-sdk/openai@npm%3A2.0.85#~/.yarn/patches/@ai-sdk-openai-npm-2.0.85-27483d1d6a.patch",
@@ -416,7 +416,8 @@
     "@langchain/openai@npm:>=0.2.0 <0.7.0": "patch:@langchain/openai@npm%3A1.0.0#~/.yarn/patches/@langchain-openai-npm-1.0.0-474d0ad9d4.patch",
     "@ai-sdk/openai@npm:^2.0.42": "patch:@ai-sdk/openai@npm%3A2.0.85#~/.yarn/patches/@ai-sdk-openai-npm-2.0.85-27483d1d6a.patch",
     "@ai-sdk/google@npm:^2.0.40": "patch:@ai-sdk/google@npm%3A2.0.40#~/.yarn/patches/@ai-sdk-google-npm-2.0.40-47e0eeee83.patch",
-    "@ai-sdk/openai-compatible@npm:^1.0.27": "patch:@ai-sdk/openai-compatible@npm%3A1.0.27#~/.yarn/patches/@ai-sdk-openai-compatible-npm-1.0.27-06f74278cf.patch"
+    "@ai-sdk/openai-compatible@npm:^1.0.27": "patch:@ai-sdk/openai-compatible@npm%3A1.0.27#~/.yarn/patches/@ai-sdk-openai-compatible-npm-1.0.27-06f74278cf.patch",
+    "@ai-sdk/google@npm:2.0.49": "patch:@ai-sdk/google@npm%3A2.0.49#~/.yarn/patches/@ai-sdk-google-npm-2.0.49-84720f41bd.patch"
   },
   "packageManager": "yarn@4.9.1",
   "lint-staged": {
diff --git a/src/renderer/src/aiCore/legacy/clients/openai/OpenAIBaseClient.ts b/src/renderer/src/aiCore/legacy/clients/openai/OpenAIBaseClient.ts
index 937827db0..9d03552cb 100644
--- a/src/renderer/src/aiCore/legacy/clients/openai/OpenAIBaseClient.ts
+++ b/src/renderer/src/aiCore/legacy/clients/openai/OpenAIBaseClient.ts
@@ -69,7 +69,7 @@ export abstract class OpenAIBaseClient<
     const sdk = await this.getSdkInstance()
     const response = (await sdk.request({
       method: 'post',
-      path: '/images/generations',
+      path: '/v1/images/generations',
       signal,
       body: {
         model,
diff --git a/src/renderer/src/aiCore/utils/reasoning.ts b/src/renderer/src/aiCore/utils/reasoning.ts
index a2364d97e..a7d602885 100644
--- a/src/renderer/src/aiCore/utils/reasoning.ts
+++ b/src/renderer/src/aiCore/utils/reasoning.ts
@@ -36,7 +36,7 @@ import {
 } from '@renderer/config/models'
 import { getStoreSetting } from '@renderer/hooks/useSettings'
 import { getAssistantSettings, getProviderByModel } from '@renderer/services/AssistantService'
-import type { Assistant, Model } from '@renderer/types'
+import type { Assistant, Model, ReasoningEffortOption } from '@renderer/types'
 import { EFFORT_RATIO, isSystemProvider, SystemProviderIds } from '@renderer/types'
 import type { OpenAIReasoningSummary } from '@renderer/types/aiCoreTypes'
 import type { ReasoningEffortOptionalParams } from '@renderer/types/sdk'
@@ -539,20 +539,25 @@ export function getAnthropicReasoningParams(
   return {}
 }
 
-// type GoogleThinkingLevel = NonNullable<GoogleGenerativeAIProviderOptions['thinkingConfig']>['thinkingLevel']
+type GoogleThinkingLevel = NonNullable<GoogleGenerativeAIProviderOptions['thinkingConfig']>['thinkingLevel']
 
-// function mapToGeminiThinkingLevel(reasoningEffort: ReasoningEffortOption): GoogelThinkingLevel {
-//   switch (reasoningEffort) {
-//     case 'low':
-//       return 'low'
-//     case 'medium':
-//       return 'medium'
-//     case 'high':
-//       return 'high'
-//     default:
-//       return 'medium'
-//   }
-// }
+function mapToGeminiThinkingLevel(reasoningEffort: ReasoningEffortOption): GoogleThinkingLevel {
+  switch (reasoningEffort) {
+    case 'default':
+      return undefined
+    case 'minimal':
+      return 'minimal'
+    case 'low':
+      return 'low'
+    case 'medium':
+      return 'medium'
+    case 'high':
+      return 'high'
+    default:
+      logger.warn('Unknown thinking level for Gemini. Fallback to medium instead.', { reasoningEffort })
+      return 'medium'
+  }
+}
 
 /**
  * 获取 Gemini 推理参数
@@ -585,15 +590,15 @@ export function getGeminiReasoningParams(
       }
     }
 
-    // TODO: 很多中转还不支持
     // https://ai.google.dev/gemini-api/docs/gemini-3?thinking=high#new_api_features_in_gemini_3
-    // if (isGemini3ThinkingTokenModel(model)) {
-    //   return {
-    //     thinkingConfig: {
-    //       thinkingLevel: mapToGeminiThinkingLevel(reasoningEffort)
-    //     }
-    //   }
-    // }
+    if (isGemini3ThinkingTokenModel(model)) {
+      return {
+        thinkingConfig: {
+          includeThoughts: true,
+          thinkingLevel: mapToGeminiThinkingLevel(reasoningEffort)
+        }
+      }
+    }
 
     const effortRatio = EFFORT_RATIO[reasoningEffort]
 
diff --git a/src/renderer/src/config/models/__tests__/reasoning.test.ts b/src/renderer/src/config/models/__tests__/reasoning.test.ts
index ba7af37f1..783cb3999 100644
--- a/src/renderer/src/config/models/__tests__/reasoning.test.ts
+++ b/src/renderer/src/config/models/__tests__/reasoning.test.ts
@@ -695,15 +695,20 @@ describe('getThinkModelType - Comprehensive Coverage', () => {
   })
 
   describe('Gemini models', () => {
-    it('should return gemini for Flash models', () => {
-      expect(getThinkModelType(createModel({ id: 'gemini-2.5-flash-latest' }))).toBe('gemini')
-      expect(getThinkModelType(createModel({ id: 'gemini-flash-latest' }))).toBe('gemini')
-      expect(getThinkModelType(createModel({ id: 'gemini-flash-lite-latest' }))).toBe('gemini')
+    it('should return gemini2_flash for Flash models', () => {
+      expect(getThinkModelType(createModel({ id: 'gemini-2.5-flash-latest' }))).toBe('gemini2_flash')
+    })
+    it('should return gemini3_flash for Gemini 3 Flash models', () => {
+      expect(getThinkModelType(createModel({ id: 'gemini-3-flash-preview' }))).toBe('gemini3_flash')
+      expect(getThinkModelType(createModel({ id: 'gemini-flash-latest' }))).toBe('gemini3_flash')
     })
 
-    it('should return gemini_pro for Pro models', () => {
-      expect(getThinkModelType(createModel({ id: 'gemini-2.5-pro-latest' }))).toBe('gemini_pro')
-      expect(getThinkModelType(createModel({ id: 'gemini-pro-latest' }))).toBe('gemini_pro')
+    it('should return gemini2_pro for Gemini 2.5 Pro models', () => {
+      expect(getThinkModelType(createModel({ id: 'gemini-2.5-pro-latest' }))).toBe('gemini2_pro')
+    })
+    it('should return gemini3_pro for Gemini 3 Pro models', () => {
+      expect(getThinkModelType(createModel({ id: 'gemini-3-pro-preview' }))).toBe('gemini3_pro')
+      expect(getThinkModelType(createModel({ id: 'gemini-pro-latest' }))).toBe('gemini3_pro')
     })
   })
 
@@ -810,7 +815,7 @@ describe('getThinkModelType - Comprehensive Coverage', () => {
             name: 'gemini-2.5-flash-latest'
           })
         )
-      ).toBe('gemini')
+      ).toBe('gemini2_flash')
     })
 
     it('should use id result when id matches', () => {
@@ -835,7 +840,7 @@ describe('getThinkModelType - Comprehensive Coverage', () => {
 
     it('should handle case insensitivity correctly', () => {
       expect(getThinkModelType(createModel({ id: 'GPT-5.1' }))).toBe('gpt5_1')
-      expect(getThinkModelType(createModel({ id: 'Gemini-2.5-Flash-Latest' }))).toBe('gemini')
+      expect(getThinkModelType(createModel({ id: 'Gemini-2.5-Flash-Latest' }))).toBe('gemini2_flash')
       expect(getThinkModelType(createModel({ id: 'DeepSeek-V3.1' }))).toBe('deepseek_hybrid')
     })
 
@@ -855,7 +860,7 @@ describe('getThinkModelType - Comprehensive Coverage', () => {
     it('should handle models with version suffixes', () => {
       expect(getThinkModelType(createModel({ id: 'gpt-5-preview-2024' }))).toBe('gpt5')
       expect(getThinkModelType(createModel({ id: 'o3-mini-2024' }))).toBe('o')
-      expect(getThinkModelType(createModel({ id: 'gemini-2.5-flash-latest-001' }))).toBe('gemini')
+      expect(getThinkModelType(createModel({ id: 'gemini-2.5-flash-latest-001' }))).toBe('gemini2_flash')
     })
 
     it('should prioritize GPT-5.1 over GPT-5 detection', () => {
@@ -955,6 +960,14 @@ describe('Gemini Models', () => {
           group: ''
         })
       ).toBe(true)
+      expect(
+        isSupportedThinkingTokenGeminiModel({
+          id: 'gemini-3-flash-preview',
+          name: '',
+          provider: '',
+          group: ''
+        })
+      ).toBe(true)
       expect(
         isSupportedThinkingTokenGeminiModel({
           id: 'google/gemini-3-pro-preview',
@@ -996,6 +1009,31 @@ describe('Gemini Models', () => {
           group: ''
         })
       ).toBe(true)
+      // Version with date suffixes
+      expect(
+        isSupportedThinkingTokenGeminiModel({
+          id: 'gemini-3-flash-preview-09-2025',
+          name: '',
+          provider: '',
+          group: ''
+        })
+      ).toBe(true)
+      expect(
+        isSupportedThinkingTokenGeminiModel({
+          id: 'gemini-3-pro-preview-09-2025',
+          name: '',
+          provider: '',
+          group: ''
+        })
+      ).toBe(true)
+      expect(
+        isSupportedThinkingTokenGeminiModel({
+          id: 'gemini-3-flash-exp-1234',
+          name: '',
+          provider: '',
+          group: ''
+        })
+      ).toBe(true)
       // Version with decimals
       expect(
         isSupportedThinkingTokenGeminiModel({
@@ -1015,7 +1053,8 @@ describe('Gemini Models', () => {
       ).toBe(true)
     })
 
-    it('should return true for gemini-3 image models', () => {
+    it('should return true for gemini-3-pro-image models only', () => {
+      // Only gemini-3-pro-image models should return true
       expect(
         isSupportedThinkingTokenGeminiModel({
           id: 'gemini-3-pro-image-preview',
@@ -1024,6 +1063,17 @@ describe('Gemini Models', () => {
           group: ''
         })
       ).toBe(true)
+      expect(
+        isSupportedThinkingTokenGeminiModel({
+          id: 'gemini-3-pro-image',
+          name: '',
+          provider: '',
+          group: ''
+        })
+      ).toBe(true)
+    })
+
+    it('should return false for other gemini-3 image models', () => {
       expect(
         isSupportedThinkingTokenGeminiModel({
           id: 'gemini-3.0-flash-image-preview',
@@ -1086,6 +1136,22 @@ describe('Gemini Models', () => {
           group: ''
         })
       ).toBe(false)
+      expect(
+        isSupportedThinkingTokenGeminiModel({
+          id: 'gemini-3-flash-preview-tts',
+          name: '',
+          provider: '',
+          group: ''
+        })
+      ).toBe(false)
+      expect(
+        isSupportedThinkingTokenGeminiModel({
+          id: 'gemini-3-pro-tts',
+          name: '',
+          provider: '',
+          group: ''
+        })
+      ).toBe(false)
     })
 
     it('should return false for older gemini models', () => {
@@ -1811,7 +1877,7 @@ describe('getModelSupportedReasoningEffortOptions', () => {
 
   describe('Gemini models', () => {
     it('should return correct options for Gemini Flash models', () => {
-      expect(getModelSupportedReasoningEffortOptions(createModel({ id: 'gemini-2.5-flash-latest' }))).toEqual([
+      expect(getModelSupportedReasoningEffortOptions(createModel({ id: 'gemini-2.5-flash' }))).toEqual([
         'default',
         'none',
         'low',
@@ -1819,36 +1885,46 @@ describe('getModelSupportedReasoningEffortOptions', () => {
         'high',
         'auto'
       ])
-      expect(getModelSupportedReasoningEffortOptions(createModel({ id: 'gemini-flash-latest' }))).toEqual([
+      expect(getModelSupportedReasoningEffortOptions(createModel({ id: 'gemini-3-flash-preview' }))).toEqual([
         'default',
-        'none',
+        'minimal',
         'low',
         'medium',
-        'high',
-        'auto'
+        'high'
+      ])
+      expect(getModelSupportedReasoningEffortOptions(createModel({ id: 'gemini-flash-latest' }))).toEqual([
+        'default',
+        'minimal',
+        'low',
+        'medium',
+        'high'
       ])
     })
 
     it('should return correct options for Gemini Pro models', () => {
-      expect(getModelSupportedReasoningEffortOptions(createModel({ id: 'gemini-2.5-pro-latest' }))).toEqual([
+      expect(getModelSupportedReasoningEffortOptions(createModel({ id: 'gemini-2.5-pro' }))).toEqual([
         'default',
         'low',
         'medium',
         'high',
         'auto'
       ])
+      expect(getModelSupportedReasoningEffortOptions(createModel({ id: 'gemini-3-pro-preview' }))).toEqual([
+        'default',
+        'low',
+        'high'
+      ])
       expect(getModelSupportedReasoningEffortOptions(createModel({ id: 'gemini-pro-latest' }))).toEqual([
         'default',
         'low',
-        'medium',
-        'high',
-        'auto'
+        'high'
       ])
     })
 
     it('should return correct options for Gemini 3 models', () => {
       expect(getModelSupportedReasoningEffortOptions(createModel({ id: 'gemini-3-flash' }))).toEqual([
         'default',
+        'minimal',
         'low',
         'medium',
         'high'
@@ -1856,7 +1932,6 @@ describe('getModelSupportedReasoningEffortOptions', () => {
       expect(getModelSupportedReasoningEffortOptions(createModel({ id: 'gemini-3-pro-preview' }))).toEqual([
         'default',
         'low',
-        'medium',
         'high'
       ])
     })
@@ -2078,7 +2153,7 @@ describe('getModelSupportedReasoningEffortOptions', () => {
 
       const geminiModel = createModel({ id: 'gemini-2.5-flash-latest' })
       const geminiResult = getModelSupportedReasoningEffortOptions(geminiModel)
-      expect(geminiResult).toEqual(MODEL_SUPPORTED_OPTIONS.gemini)
+      expect(geminiResult).toEqual(MODEL_SUPPORTED_OPTIONS.gemini2_flash)
     })
   })
 })
diff --git a/src/renderer/src/config/models/__tests__/utils.test.ts b/src/renderer/src/config/models/__tests__/utils.test.ts
index 56a9e6041..0b2294fd4 100644
--- a/src/renderer/src/config/models/__tests__/utils.test.ts
+++ b/src/renderer/src/config/models/__tests__/utils.test.ts
@@ -21,6 +21,8 @@ import {
   getModelSupportedVerbosity,
   groupQwenModels,
   isAnthropicModel,
+  isGemini3FlashModel,
+  isGemini3ProModel,
   isGeminiModel,
   isGemmaModel,
   isGenerateImageModels,
@@ -462,6 +464,101 @@ describe('model utils', () => {
       })
     })
 
+    describe('isGemini3FlashModel', () => {
+      it('detects gemini-3-flash model', () => {
+        expect(isGemini3FlashModel(createModel({ id: 'gemini-3-flash' }))).toBe(true)
+      })
+
+      it('detects gemini-3-flash-preview model', () => {
+        expect(isGemini3FlashModel(createModel({ id: 'gemini-3-flash-preview' }))).toBe(true)
+      })
+
+      it('detects gemini-3-flash with version suffixes', () => {
+        expect(isGemini3FlashModel(createModel({ id: 'gemini-3-flash-latest' }))).toBe(true)
+        expect(isGemini3FlashModel(createModel({ id: 'gemini-3-flash-preview-09-2025' }))).toBe(true)
+        expect(isGemini3FlashModel(createModel({ id: 'gemini-3-flash-exp-1234' }))).toBe(true)
+      })
+
+      it('detects gemini-flash-latest alias', () => {
+        expect(isGemini3FlashModel(createModel({ id: 'gemini-flash-latest' }))).toBe(true)
+        expect(isGemini3FlashModel(createModel({ id: 'Gemini-Flash-Latest' }))).toBe(true)
+      })
+
+      it('detects gemini-3-flash with uppercase', () => {
+        expect(isGemini3FlashModel(createModel({ id: 'Gemini-3-Flash' }))).toBe(true)
+        expect(isGemini3FlashModel(createModel({ id: 'GEMINI-3-FLASH-PREVIEW' }))).toBe(true)
+      })
+
+      it('excludes gemini-3-flash-image models', () => {
+        expect(isGemini3FlashModel(createModel({ id: 'gemini-3-flash-image-preview' }))).toBe(false)
+        expect(isGemini3FlashModel(createModel({ id: 'gemini-3-flash-image' }))).toBe(false)
+      })
+
+      it('returns false for non-flash gemini-3 models', () => {
+        expect(isGemini3FlashModel(createModel({ id: 'gemini-3-pro' }))).toBe(false)
+        expect(isGemini3FlashModel(createModel({ id: 'gemini-3-pro-preview' }))).toBe(false)
+        expect(isGemini3FlashModel(createModel({ id: 'gemini-3-pro-image-preview' }))).toBe(false)
+      })
+
+      it('returns false for other gemini models', () => {
+        expect(isGemini3FlashModel(createModel({ id: 'gemini-2-flash' }))).toBe(false)
+        expect(isGemini3FlashModel(createModel({ id: 'gemini-2-flash-preview' }))).toBe(false)
+        expect(isGemini3FlashModel(createModel({ id: 'gemini-2.5-flash-preview-09-2025' }))).toBe(false)
+      })
+
+      it('returns false for null/undefined models', () => {
+        expect(isGemini3FlashModel(null)).toBe(false)
+        expect(isGemini3FlashModel(undefined)).toBe(false)
+      })
+    })
+
+    describe('isGemini3ProModel', () => {
+      it('detects gemini-3-pro model', () => {
+        expect(isGemini3ProModel(createModel({ id: 'gemini-3-pro' }))).toBe(true)
+      })
+
+      it('detects gemini-3-pro-preview model', () => {
+        expect(isGemini3ProModel(createModel({ id: 'gemini-3-pro-preview' }))).toBe(true)
+      })
+
+      it('detects gemini-3-pro with version suffixes', () => {
+        expect(isGemini3ProModel(createModel({ id: 'gemini-3-pro-latest' }))).toBe(true)
+        expect(isGemini3ProModel(createModel({ id: 'gemini-3-pro-preview-09-2025' }))).toBe(true)
+        expect(isGemini3ProModel(createModel({ id: 'gemini-3-pro-exp-1234' }))).toBe(true)
+      })
+
+      it('detects gemini-pro-latest alias', () => {
+        expect(isGemini3ProModel(createModel({ id: 'gemini-pro-latest' }))).toBe(true)
+        expect(isGemini3ProModel(createModel({ id: 'Gemini-Pro-Latest' }))).toBe(true)
+      })
+
+      it('detects gemini-3-pro with uppercase', () => {
+        expect(isGemini3ProModel(createModel({ id: 'Gemini-3-Pro' }))).toBe(true)
+        expect(isGemini3ProModel(createModel({ id: 'GEMINI-3-PRO-PREVIEW' }))).toBe(true)
+      })
+
+      it('excludes gemini-3-pro-image models', () => {
+        expect(isGemini3ProModel(createModel({ id: 'gemini-3-pro-image-preview' }))).toBe(false)
+        expect(isGemini3ProModel(createModel({ id: 'gemini-3-pro-image' }))).toBe(false)
+        expect(isGemini3ProModel(createModel({ id: 'gemini-3-pro-image-latest' }))).toBe(false)
+      })
+
+      it('returns false for non-pro gemini-3 models', () => {
+        expect(isGemini3ProModel(createModel({ id: 'gemini-3-flash' }))).toBe(false)
+        expect(isGemini3ProModel(createModel({ id: 'gemini-3-flash-preview' }))).toBe(false)
+      })
+
+      it('returns false for other gemini models', () => {
+        expect(isGemini3ProModel(createModel({ id: 'gemini-2-pro' }))).toBe(false)
+        expect(isGemini3ProModel(createModel({ id: 'gemini-2.5-pro-preview-09-2025' }))).toBe(false)
+      })
+
+      it('returns false for null/undefined models', () => {
+        expect(isGemini3ProModel(null)).toBe(false)
+        expect(isGemini3ProModel(undefined)).toBe(false)
+      })
+    })
+
     describe('isZhipuModel', () => {
       it('detects Zhipu models by provider', () => {
         expect(isZhipuModel(createModel({ provider: 'zhipu' }))).toBe(true)
diff --git a/src/renderer/src/config/models/reasoning.ts b/src/renderer/src/config/models/reasoning.ts
index faa04721e..ea7d8747f 100644
--- a/src/renderer/src/config/models/reasoning.ts
+++ b/src/renderer/src/config/models/reasoning.ts
@@ -20,7 +20,7 @@ import {
   isOpenAIReasoningModel,
   isSupportedReasoningEffortOpenAIModel
 } from './openai'
-import { GEMINI_FLASH_MODEL_REGEX, isGemini3ThinkingTokenModel } from './utils'
+import { GEMINI_FLASH_MODEL_REGEX, isGemini3FlashModel, isGemini3ProModel } from './utils'
 import { isTextToImageModel } from './vision'
 
 // Reasoning models
@@ -43,9 +43,10 @@ export const MODEL_SUPPORTED_REASONING_EFFORT = {
   gpt52pro: ['medium', 'high', 'xhigh'] as const,
   grok: ['low', 'high'] as const,
   grok4_fast: ['auto'] as const,
-  gemini: ['low', 'medium', 'high', 'auto'] as const,
-  gemini3: ['low', 'medium', 'high'] as const,
-  gemini_pro: ['low', 'medium', 'high', 'auto'] as const,
+  gemini2_flash: ['low', 'medium', 'high', 'auto'] as const,
+  gemini2_pro: ['low', 'medium', 'high', 'auto'] as const,
+  gemini3_flash: ['minimal', 'low', 'medium', 'high'] as const,
+  gemini3_pro: ['low', 'high'] as const,
   qwen: ['low', 'medium', 'high'] as const,
   qwen_thinking: ['low', 'medium', 'high'] as const,
   doubao: ['auto', 'high'] as const,
@@ -53,6 +54,7 @@ export const MODEL_SUPPORTED_REASONING_EFFORT = {
   doubao_after_251015: ['minimal', 'low', 'medium', 'high'] as const,
   hunyuan: ['auto'] as const,
   mimo: ['auto'] as const,
+  mimo: ['auto'] as const,
   zhipu: ['auto'] as const,
   perplexity: ['low', 'medium', 'high'] as const,
   deepseek_hybrid: ['auto'] as const
@@ -73,9 +75,10 @@ export const MODEL_SUPPORTED_OPTIONS: ThinkingOptionConfig = {
   gpt52pro: ['default', ...MODEL_SUPPORTED_REASONING_EFFORT.gpt52pro] as const,
   grok: ['default', ...MODEL_SUPPORTED_REASONING_EFFORT.grok] as const,
   grok4_fast: ['default', 'none', ...MODEL_SUPPORTED_REASONING_EFFORT.grok4_fast] as const,
-  gemini: ['default', 'none', ...MODEL_SUPPORTED_REASONING_EFFORT.gemini] as const,
-  gemini_pro: ['default', ...MODEL_SUPPORTED_REASONING_EFFORT.gemini_pro] as const,
-  gemini3: ['default', ...MODEL_SUPPORTED_REASONING_EFFORT.gemini3] as const,
+  gemini2_flash: ['default', 'none', ...MODEL_SUPPORTED_REASONING_EFFORT.gemini2_flash] as const,
+  gemini2_pro: ['default', ...MODEL_SUPPORTED_REASONING_EFFORT.gemini2_pro] as const,
+  gemini3_flash: ['default', ...MODEL_SUPPORTED_REASONING_EFFORT.gemini3_flash] as const,
+  gemini3_pro: ['default', ...MODEL_SUPPORTED_REASONING_EFFORT.gemini3_pro] as const,
   qwen: ['default', 'none', ...MODEL_SUPPORTED_REASONING_EFFORT.qwen] as const,
   qwen_thinking: ['default', ...MODEL_SUPPORTED_REASONING_EFFORT.qwen_thinking] as const,
   doubao: ['default', 'none', ...MODEL_SUPPORTED_REASONING_EFFORT.doubao] as const,
@@ -102,8 +105,7 @@ const _getThinkModelType = (model: Model): ThinkingModelType => {
   const modelId = getLowerBaseModelName(model.id)
   if (isOpenAIDeepResearchModel(model)) {
     return 'openai_deep_research'
-  }
-  if (isGPT51SeriesModel(model)) {
+  } else if (isGPT51SeriesModel(model)) {
     if (modelId.includes('codex')) {
       thinkingModelType = 'gpt5_1_codex'
       if (isGPT51CodexMaxModel(model)) {
@@ -131,16 +133,18 @@ const _getThinkModelType = (model: Model): ThinkingModelType => {
   } else if (isGrok4FastReasoningModel(model)) {
     thinkingModelType = 'grok4_fast'
   } else if (isSupportedThinkingTokenGeminiModel(model)) {
-    if (GEMINI_FLASH_MODEL_REGEX.test(model.id)) {
-      thinkingModelType = 'gemini'
+    if (isGemini3FlashModel(model)) {
+      thinkingModelType = 'gemini3_flash'
+    } else if (isGemini3ProModel(model)) {
+      thinkingModelType = 'gemini3_pro'
+    } else if (GEMINI_FLASH_MODEL_REGEX.test(model.id)) {
+      thinkingModelType = 'gemini2_flash'
     } else {
-      thinkingModelType = 'gemini_pro'
+      thinkingModelType = 'gemini2_pro'
     }
-    if (isGemini3ThinkingTokenModel(model)) {
-      thinkingModelType = 'gemini3'
-    }
-  } else if (isSupportedReasoningEffortGrokModel(model)) thinkingModelType = 'grok'
-  else if (isSupportedThinkingTokenQwenModel(model)) {
+  } else if (isSupportedReasoningEffortGrokModel(model)) {
+    thinkingModelType = 'grok'
+  } else if (isSupportedThinkingTokenQwenModel(model)) {
     if (isQwenAlwaysThinkModel(model)) {
       thinkingModelType = 'qwen_thinking'
     }
@@ -153,11 +157,17 @@ const _getThinkModelType = (model: Model): ThinkingModelType => {
     } else {
       thinkingModelType = 'doubao_no_auto'
     }
-  } else if (isSupportedThinkingTokenHunyuanModel(model)) thinkingModelType = 'hunyuan'
-  else if (isSupportedReasoningEffortPerplexityModel(model)) thinkingModelType = 'perplexity'
-  else if (isSupportedThinkingTokenZhipuModel(model)) thinkingModelType = 'zhipu'
-  else if (isDeepSeekHybridInferenceModel(model)) thinkingModelType = 'deepseek_hybrid'
-  else if (isSupportedThinkingTokenMiMoModel(model)) thinkingModelType = 'mimo'
+  } else if (isSupportedThinkingTokenHunyuanModel(model)) {
+    thinkingModelType = 'hunyuan'
+  } else if (isSupportedReasoningEffortPerplexityModel(model)) {
+    thinkingModelType = 'perplexity'
+  } else if (isSupportedThinkingTokenZhipuModel(model)) {
+    thinkingModelType = 'zhipu'
+  } else if (isDeepSeekHybridInferenceModel(model)) {
+    thinkingModelType = 'deepseek_hybrid'
+  } else if (isSupportedThinkingTokenMiMoModel(model)) {
+    thinkingModelType = 'mimo'
+  }
   return thinkingModelType
 }
 
diff --git a/src/renderer/src/config/models/utils.ts b/src/renderer/src/config/models/utils.ts
index d12680d57..0a60244e1 100644
--- a/src/renderer/src/config/models/utils.ts
+++ b/src/renderer/src/config/models/utils.ts
@@ -282,3 +282,43 @@ export const isGemini3ThinkingTokenModel = (model: Model) => {
   const modelId = getLowerBaseModelName(model.id)
   return isGemini3Model(model) && !modelId.includes('image')
 }
+
+/**
+ * Check if the model is a Gemini 3 Flash model
+ * Matches: gemini-3-flash, gemini-3-flash-preview, gemini-3-flash-preview-09-2025, gemini-flash-latest (alias)
+ * Excludes: gemini-3-flash-image-preview
+ * @param model - The model to check
+ * @returns true if the model is a Gemini 3 Flash model
+ */
+export const isGemini3FlashModel = (model: Model | undefined | null): boolean => {
+  if (!model) {
+    return false
+  }
+  const modelId = getLowerBaseModelName(model.id)
+  // Check for gemini-flash-latest alias (currently points to gemini-3-flash, may change in future)
+  if (modelId === 'gemini-flash-latest') {
+    return true
+  }
+  // Check for gemini-3-flash with optional suffixes, excluding image variants
+  return /gemini-3-flash(?!-image)(?:-[\w-]+)*$/i.test(modelId)
+}
+
+/**
+ * Check if the model is a Gemini 3 Pro model
+ * Matches: gemini-3-pro, gemini-3-pro-preview, gemini-3-pro-preview-09-2025, gemini-pro-latest (alias)
+ * Excludes: gemini-3-pro-image-preview
+ * @param model - The model to check
+ * @returns true if the model is a Gemini 3 Pro model
+ */
+export const isGemini3ProModel = (model: Model | undefined | null): boolean => {
+  if (!model) {
+    return false
+  }
+  const modelId = getLowerBaseModelName(model.id)
+  // Check for gemini-pro-latest alias (currently points to gemini-3-pro, may change in future)
+  if (modelId === 'gemini-pro-latest') {
+    return true
+  }
+  // Check for gemini-3-pro with optional suffixes, excluding image variants
+  return /gemini-3-pro(?!-image)(?:-[\w-]+)*$/i.test(modelId)
+}
diff --git a/src/renderer/src/i18n/locales/zh-cn.json b/src/renderer/src/i18n/locales/zh-cn.json
index f0d4adf4c..0e5b2f60e 100644
--- a/src/renderer/src/i18n/locales/zh-cn.json
+++ b/src/renderer/src/i18n/locales/zh-cn.json
@@ -560,7 +560,7 @@
         "medium": "斟酌",
         "medium_description": "中强度推理",
         "minimal": "微念",
-        "minimal_description": "最小程度的思考",
+        "minimal_description": "最小程度的推理",
         "off": "关闭",
         "off_description": "禁用推理",
         "xhigh": "穷究",
diff --git a/src/renderer/src/types/index.ts b/src/renderer/src/types/index.ts
index 67a79f656..5893a31ba 100644
--- a/src/renderer/src/types/index.ts
+++ b/src/renderer/src/types/index.ts
@@ -95,9 +95,10 @@ const ThinkModelTypes = [
   'gpt52pro',
   'grok',
   'grok4_fast',
-  'gemini',
-  'gemini_pro',
-  'gemini3',
+  'gemini2_flash',
+  'gemini2_pro',
+  'gemini3_flash',
+  'gemini3_pro',
   'qwen',
   'qwen_thinking',
   'doubao',
diff --git a/yarn.lock b/yarn.lock
index ab781d370..d9d5ec1d6 100644
--- a/yarn.lock
+++ b/yarn.lock
@@ -102,6 +102,18 @@ __metadata:
   languageName: node
   linkType: hard
 
+"@ai-sdk/anthropic@npm:2.0.56":
+  version: 2.0.56
+  resolution: "@ai-sdk/anthropic@npm:2.0.56"
+  dependencies:
+    "@ai-sdk/provider": "npm:2.0.0"
+    "@ai-sdk/provider-utils": "npm:3.0.19"
+  peerDependencies:
+    zod: ^3.25.76 || ^4.1.8
+  checksum: 10c0/f2b6029c92443f831a2d124420e805d057668003067b1f677a4292d02f27aa3ad533374ea996d77ede7746a42c46fb94a8f2d8c0e7758a4555ea18c8b532052c
+  languageName: node
+  linkType: hard
+
 "@ai-sdk/azure@npm:^2.0.87":
   version: 2.0.87
   resolution: "@ai-sdk/azure@npm:2.0.87"
@@ -166,42 +178,42 @@ __metadata:
   languageName: node
   linkType: hard
 
-"@ai-sdk/google-vertex@npm:^3.0.79":
-  version: 3.0.79
-  resolution: "@ai-sdk/google-vertex@npm:3.0.79"
+"@ai-sdk/google-vertex@npm:^3.0.94":
+  version: 3.0.94
+  resolution: "@ai-sdk/google-vertex@npm:3.0.94"
   dependencies:
-    "@ai-sdk/anthropic": "npm:2.0.49"
-    "@ai-sdk/google": "npm:2.0.43"
+    "@ai-sdk/anthropic": "npm:2.0.56"
+    "@ai-sdk/google": "npm:2.0.49"
     "@ai-sdk/provider": "npm:2.0.0"
-    "@ai-sdk/provider-utils": "npm:3.0.17"
-    google-auth-library: "npm:^9.15.0"
+    "@ai-sdk/provider-utils": "npm:3.0.19"
+    google-auth-library: "npm:^10.5.0"
   peerDependencies:
     zod: ^3.25.76 || ^4.1.8
-  checksum: 10c0/a86949b8d4a855409acdf7dc8d93ad9ea8ccf2bc3849acbe1ecbe4d6d66f06bcb5242f0df8eea24214e78732618b71ec8a019cbbeab16366f9ad3c860c5d8d30
+  checksum: 10c0/68e2ee9e6525a5e43f90304980e64bf2a4227fd3ce74a7bf17e5ace094ea1bca8f8f18a8cc332a492fee4b912568a768f7479a4eed8148b84e7de1adf4104ad0
   languageName: node
   linkType: hard
 
-"@ai-sdk/google@npm:2.0.43":
-  version: 2.0.43
-  resolution: "@ai-sdk/google@npm:2.0.43"
+"@ai-sdk/google@npm:2.0.49":
+  version: 2.0.49
+  resolution: "@ai-sdk/google@npm:2.0.49"
   dependencies:
     "@ai-sdk/provider": "npm:2.0.0"
-    "@ai-sdk/provider-utils": "npm:3.0.17"
+    "@ai-sdk/provider-utils": "npm:3.0.19"
   peerDependencies:
     zod: ^3.25.76 || ^4.1.8
-  checksum: 10c0/5a421a9746cf8cbdf3bb7fb49426453a4fe0e354ea55a0123e628afb7acf9bb19959d512c0f8e6d7dbefbfa7e1cef4502fc146149007258a8eeb57743ac5e9e5
+  checksum: 10c0/f3f8acfcd956edc7d807d22963d5eff0f765418f1f2c7d18615955ccdfcebb4d43cc26ce1f712c6a53572f1d8becc0773311b77b1f1bf1af87d675c5f017d5a4
   languageName: node
   linkType: hard
 
-"@ai-sdk/google@patch:@ai-sdk/google@npm%3A2.0.43#~/.yarn/patches/@ai-sdk-google-npm-2.0.43-689ed559b3.patch":
-  version: 2.0.43
-  resolution: "@ai-sdk/google@patch:@ai-sdk/google@npm%3A2.0.43#~/.yarn/patches/@ai-sdk-google-npm-2.0.43-689ed559b3.patch::version=2.0.43&hash=4dde1e"
+"@ai-sdk/google@patch:@ai-sdk/google@npm%3A2.0.49#~/.yarn/patches/@ai-sdk-google-npm-2.0.49-84720f41bd.patch":
+  version: 2.0.49
+  resolution: "@ai-sdk/google@patch:@ai-sdk/google@npm%3A2.0.49#~/.yarn/patches/@ai-sdk-google-npm-2.0.49-84720f41bd.patch::version=2.0.49&hash=406c25"
   dependencies:
     "@ai-sdk/provider": "npm:2.0.0"
-    "@ai-sdk/provider-utils": "npm:3.0.17"
+    "@ai-sdk/provider-utils": "npm:3.0.19"
   peerDependencies:
     zod: ^3.25.76 || ^4.1.8
-  checksum: 10c0/4cfd17e9c47f2b742d8a0b1ca3532b4dc48753088363b74b01a042f63652174fa9a3fbf655a23f823974c673121dffbd2d192bb0c1bf158da4e2bf498fc76527
+  checksum: 10c0/8d4d881583c2301dce8a4e3066af2ba7d99b30520b6219811f90271c93bf8a07dc23e752fa25ffd0e72c6ec56e97d40d32e04072a362accf7d01a745a2d2a352
   languageName: node
   linkType: hard
 
@@ -10051,8 +10063,8 @@ __metadata:
     "@ai-sdk/anthropic": "npm:^2.0.49"
     "@ai-sdk/cerebras": "npm:^1.0.31"
     "@ai-sdk/gateway": "npm:^2.0.15"
-    "@ai-sdk/google": "patch:@ai-sdk/google@npm%3A2.0.43#~/.yarn/patches/@ai-sdk-google-npm-2.0.43-689ed559b3.patch"
-    "@ai-sdk/google-vertex": "npm:^3.0.79"
+    "@ai-sdk/google": "patch:@ai-sdk/google@npm%3A2.0.49#~/.yarn/patches/@ai-sdk-google-npm-2.0.49-84720f41bd.patch"
+    "@ai-sdk/google-vertex": "npm:^3.0.94"
     "@ai-sdk/huggingface": "npm:^0.0.10"
     "@ai-sdk/mistral": "npm:^2.0.24"
     "@ai-sdk/openai": "patch:@ai-sdk/openai@npm%3A2.0.85#~/.yarn/patches/@ai-sdk-openai-npm-2.0.85-27483d1d6a.patch"
@@ -15499,6 +15511,18 @@ __metadata:
   languageName: node
   linkType: hard
 
+"gaxios@npm:^7.0.0":
+  version: 7.1.3
+  resolution: "gaxios@npm:7.1.3"
+  dependencies:
+    extend: "npm:^3.0.2"
+    https-proxy-agent: "npm:^7.0.1"
+    node-fetch: "npm:^3.3.2"
+    rimraf: "npm:^5.0.1"
+  checksum: 10c0/a4a1cdf9a392c0c22e9734a40dca5a77a2903f505b939a50f1e68e312458b1289b7993d2f72d011426e89657cae77a3aa9fc62fb140e8ba90a1faa31fdbde4d2
+  languageName: node
+  linkType: hard
+
 "gcp-metadata@npm:^6.1.0":
   version: 6.1.1
   resolution: "gcp-metadata@npm:6.1.1"
@@ -15510,6 +15534,17 @@ __metadata:
   languageName: node
   linkType: hard
 
+"gcp-metadata@npm:^8.0.0":
+  version: 8.1.2
+  resolution: "gcp-metadata@npm:8.1.2"
+  dependencies:
+    gaxios: "npm:^7.0.0"
+    google-logging-utils: "npm:^1.0.0"
+    json-bigint: "npm:^1.0.0"
+  checksum: 10c0/15a61231a9410dc11c2828d2c9fdc8b0a939f1af746195c44edc6f2ffea0acab52cef3a7b9828069a36fd5d68bda730f7328a415fe42a01258f6e249dfba6908
+  languageName: node
+  linkType: hard
+
 "gensync@npm:^1.0.0-beta.2":
   version: 1.0.0-beta.2
   resolution: "gensync@npm:1.0.0-beta.2"
@@ -15733,7 +15768,22 @@ __metadata:
   languageName: node
   linkType: hard
 
-"google-auth-library@npm:^9.14.2, google-auth-library@npm:^9.15.0, google-auth-library@npm:^9.15.1, google-auth-library@npm:^9.4.2":
+"google-auth-library@npm:^10.5.0":
+  version: 10.5.0
+  resolution: "google-auth-library@npm:10.5.0"
+  dependencies:
+    base64-js: "npm:^1.3.0"
+    ecdsa-sig-formatter: "npm:^1.0.11"
+    gaxios: "npm:^7.0.0"
+    gcp-metadata: "npm:^8.0.0"
+    google-logging-utils: "npm:^1.0.0"
+    gtoken: "npm:^8.0.0"
+    jws: "npm:^4.0.0"
+  checksum: 10c0/49d3931d20b1f4a4d075216bf5518e2b3396dcf441a8f1952611cf3b6080afb1261c3d32009609047ee4a1cc545269a74b4957e6bba9cce840581df309c4b145
+  languageName: node
+  linkType: hard
+
+"google-auth-library@npm:^9.14.2, google-auth-library@npm:^9.15.1, google-auth-library@npm:^9.4.2":
   version: 9.15.1
   resolution: "google-auth-library@npm:9.15.1"
   dependencies:
@@ -15754,6 +15804,13 @@ __metadata:
   languageName: node
   linkType: hard
 
+"google-logging-utils@npm:^1.0.0":
+  version: 1.1.3
+  resolution: "google-logging-utils@npm:1.1.3"
+  checksum: 10c0/e65201c7e96543bd1423b9324013736646b9eed60941e0bfa47b9bfd146d2f09cf3df1c99ca60b7d80a726075263ead049ee72de53372cb8458c3bc55c2c1e59
+  languageName: node
+  linkType: hard
+
 "gopd@npm:^1.0.1, gopd@npm:^1.2.0":
   version: 1.2.0
   resolution: "gopd@npm:1.2.0"
@@ -15842,6 +15899,16 @@ __metadata:
   languageName: node
   linkType: hard
 
+"gtoken@npm:^8.0.0":
+  version: 8.0.0
+  resolution: "gtoken@npm:8.0.0"
+  dependencies:
+    gaxios: "npm:^7.0.0"
+    jws: "npm:^4.0.0"
+  checksum: 10c0/058538e5bbe081d30ada5f1fd34d3a8194357c2e6ecbf7c8a98daeefbf13f7e06c15649c7dace6a1d4cc3bc6dc5483bd484d6d7adc5852021896d7c05c439f37
+  languageName: node
+  linkType: hard
+
 "hachure-fill@npm:^0.5.2":
   version: 0.5.2
   resolution: "hachure-fill@npm:0.5.2"
@@ -22778,7 +22845,7 @@ __metadata:
   languageName: node
   linkType: hard
 
-"rimraf@npm:^5.0.10":
+"rimraf@npm:^5.0.1, rimraf@npm:^5.0.10":
   version: 5.0.10
   resolution: "rimraf@npm:5.0.10"
   dependencies: